Massive multi-type 3D datasets in parallel Environments

Most existing software for handling of 3D data-sets is based on historic levels of computer performance, and programmed according to a sequential paradigm. Low cost parallel processing hardware is now available due to the recent introduction of multi-core CPUs and programmable graphics cards. The handling and processing of large 3D data-sets is computational expensive and will benefit considerably from exploiting such parallel resources. Parallel3D addresses the emerging and the expected future growth i n parallel computing power of commodity hardware to deal with performance and quality challenges related to large multi-source and multi-type 3D data-sets. The size of such data-sets is rapidly increasing due better methods for gathering information. A si milar increase takes place in data-sets produced by computers e.g. results from simulation programs. Three research challenges related to large 3D data-sets and parallel environments are addressed: - Architecture: Scene graphs in a parallel environment; - Multi-resolution representation of 3D data for reduced data size and reduced bandwidth requirements; - Volume visualization of large out-of-core data sets. The company Systems-in-Motion (SIM) provides part of the project funding, and will cooperate cl osely with the project to ensure that the project results are relevant for SIM and industry. SIM will provide program codes to avoid unnecessary programming of support software. The project will demonstrate the developed prototype toolkits on industrial t ype data-sets. The research partners are SINTEF ICT and the Center of Mathematics for Applications at the University of Oslo. The project will supervise master students and two PhD fellows. We believe that in many branches of public the sector, industry and commerce improved tools for handling large 3D data-sets will help decision making and consequently improve commercial and/or operational success.